[Bug] C++ API and core tests fail with "assert 3 == 0" across all GPUs and CUDA 13.x versions
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 3k
- Forks
- 410
- Avg merge
- 3d 18h
- Merged PRs (30d)
- 78
Description
Bug Description
C++ test suites //tests/core:core_tests and //tests:cpp_api_tests consistently fail with assert 3 == 0 across all tested GPU architectures (RTX 3070, B100-TS2, H100, A100) and all CUDA 13.x versions. The failure is 100% reproducible across every configuration, suggesting a systemic issue rather than a hardware- or driver-specific one.
Environment
- GPUs: RTX 3070, B100-TS2, H100, A100
- Arch: x86_64
- CUDA: 13.2.0 / 13.1.1 / 13.0.2
- OS: Ubuntu 24.04
- cuDNN: 8.9.7.29
- Myelin: 2.17.78+7
- CASK: 5.16.17+1
- Python: 3.12
- Package: qa_tar_py3.12
Failing Tests
FAILED test_torch_trt_cpp[//tests/core:core_tests] - assert 3 == 0
FAILED test_torch_trt_cpp[//tests:cpp_api_tests] - assert 3 == 0
Reproducible Configurations
| GPU | CUDA | Result |
|---|---|---|
| RTX 3070/x86_64 | r13.0.2, r13.1.1, r13.2.0 | FAILED |
| A100/x86_64 | r13.0.2, r13.1.1, r13.2.0 | FAILED |
| H100/x86_64 | r13.1.1, r13.2.0 | FAILED |
| B100-TS2/x86_64 | r13.0.2, r13.1.1, r13.2.0 | FAILED |
All 22 logged test runs failed — no passing configuration observed.
Steps to Reproduce
- Run on any of the listed GPUs with CUDA 13.x and the environment above
- Execute the C++ test suites:
//tests/core:core_tests//tests:cpp_api_tests
Expected Behavior
Both C++ test suites should pass with assert 0 == 0.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by running //tests/core:core_tests and //tests:cpp_api_tests in the listed CUDA 13.x environment, then trace where the assert 3 == 0 failure is reported. Compare the failing behavior across the provided GPU and CUDA configurations; done means both C++ test suites pass with the expected zero assertion.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp, python
- Domain
- machine-learning, testing-qa
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100